Use progress bar in tqdm

[cphyc]

This PR add the possibility to have the progress bar using tqdm (https://github.com/noamraph/tqdm) to show the progress as asked in #176 .

[basnijholt]

I am just leaving this snippet here for anyone in the future to Google tqdm and ipyparallel in the future (like I did now). This is a non-blocking way of doing so:

def live_info(ar, *, update_interval=0.1):
    import asyncio
    from tqdm import tqdm_notebook
    pbar = tqdm_notebook(total=len(ar.msg_ids))
    async def update():
        val = 0
        while not ar.ready():
            new_val = ar.progress
            pbar.update(new_val - val)
            val = new_val
            await asyncio.sleep(update_interval)
        pbar.update(pbar.total - pbar.n)

    asyncio.get_event_loop().create_task(update())

[minrk]

Woo, merged after only four years. Sorry for the wait, @cphyc!

[GF-Huang]

The progress bar seems based on task number (CPU count), how to make it base on the sequences/collection/iteration while using DirectView? I know LoadBalancedView works, but due to #539 I can't use LoadBalancedView.

For example I have a list: items = range(100000), how to make the progress bar count based on 100000 rather than 8 (my computer CPU count).

[minrk]

The progress bar is based on the number of 'tasks' (messages). IPython doesn't have any view into fractions of messages. In a DirectView, that's always one per engine. In a LoadBalancedView, that's one per item by default, but controllable via the chunksize argument.